README.md

---
license: mit
library_name: transformers
pipeline_tag: image-text-to-text
tags:
- reinforcement-learning
- agent
- gui-agent
- vl-model
---
---

# RLAnything: Forge Environment, Policy, and Reward Model in Completely Dynamic RL System

[Paper](https://arxiv.org/abs/2602.02488) | [Code](https://github.com/Gen-Verse/Open-AgentRL) | [Blog](https://yinjjiew.github.io/projects/rlanything/)

**RLAnything** is a reinforcement learning framework that dynamically forges environment, policy, and reward models through closed-loop optimization, amplifying learning signals and strengthening the overall RL system for any LLM or agentic scenarios.

### Highlights

*   **Integrated Feedback for Policy:** The policy is trained with integrated outcome and step-wise signals from the reward model, outperforming traditional outcome-only signals.
*   **Consistency Feedback for Reward Model:** The reward model is jointly optimized via consistency feedback, which in turn further improves policy training.
*   **Critic Feedback for Environment:** Theory-motivated automatic environment adaptation improves training for both the reward and policy models by leveraging critic feedback from each, enabling learning from experience.

<p align="center">
  <img src="https://github.com/yinjjiew/Data/raw/main/rlanything/rlanythingoverview.png" width="100%"/>
</p>

### Performance

RLAnything yields substantial gains across various representative LLM and agentic tasks, boosting Qwen3-VL-8B-Thinking by 9.1% on OSWorld and Qwen2.5-7B-Instruct by 18.7% and 11.9% on AlfWorld and LiveBench, respectively.

<p align="center">
  <img src="https://github.com/yinjjiew/Data/raw/main/rlanything/rlanythingscaleosworld.png" width="70%"/>
</p>


<p align="center">
  <img src="https://github.com/yinjjiew/Data/raw/main/rlanything/rlanythingosworldbench.png" width="100%"/>
</p>


## Citation

```bibtex
@article{wang2026rlanything,
  title={RLAnything: Forge Environment, Policy, and Reward Model in Completely Dynamic RL System},
  author={Wang, Yinjie and Xie, Tianbao and Shen, Ke and Wang, Mengdi and Yang, Ling},
  journal={arXiv preprint arXiv:2602.02488},
  year={2026}
}
```